I started out in a post room in 1995 and taught myself everything since. These days I design and build AI products, from the first idea to the finished thing.

My first job was sorting post and looking after the franking machines. After that I ran overnight backups in a computer room, on UNIX, OS/2 and mainframe systems, and checked the restores actually worked. Then came a helpdesk, then second-line support, then third-line, on clinical software used by NHS walk-in centres, NHS Direct and NHS 24, as well as sites in Australia, Spain, Portugal and Norway. That was about three thousand clients, with a clustered SQL Server behind the lot.

In 2012 I moved into Capita's development team to look after software packaging and build automation. That's where I started writing production code, mostly PowerShell and JavaScript to automate builds, deployments and systems jobs. In time I ended up running the whole Azure DevOps setup and the CI/CD pipelines every product team relied on.

I don't have a computer science degree and never have. What I do have is a lot of years as the person who had to make things work, and a habit of learning whatever the next problem needed.

In 2020 I started publishing commercial JavaScript plugins for the Bubble platform. There are 47 of them now, covering things like video players and recorders, 2D and 3D maps, service workers, rich text editing, real-time video, QR scanning and encryption. They still bring in money every month, and most of my client work came from people who'd found them.

Over time the plugins moved more and more towards AI: an agent runtime, RAG, agentic tools, real-time speech over WebRTC, LiveKit, live avatars and speech animation. That's the same area I work in full time now, and the plugins are where I learned most of it, with paying customers reporting the bugs.

Alongside that, I've built eight products of my own, Vantage City OS, Lookout, Glass Box, Loose Brief, Interactive Avatar, Rival Radar, Chart Chic and Shard Islands, and all eight are live. I left Capita in July 2026 to do this full time.

I use AI coding agents as my main tool, and they've changed how much one person can build. They only work well if someone is making the real decisions, though. I plan the architecture, the database and the security myself, the agent writes code to that plan, and I check every change against real data before it goes live.

The biggest risk with AI is that it makes things up and sounds sure of itself, so I design around that. In Rival Radar, for example, the AI never gets to suggest competitors from memory. It only works with what turns up in real searches.

I also try to keep up with what's new. I subscribe to ByteSize, Tech Brew, The Batch and The Rundown AI, and most weeks something in one of them ends up with me spending an afternoon trying it out.

Building AI-powered websites and products for clients and customers. It's the work I enjoy most and what I've been doing independently for the last few years, from the first idea through to the finished, live product.

If I had to describe how I work, I'd say creative technologist: I'll have an idea in the morning, work out how it should function, and have something running by the evening. On job boards it's usually called AI engineer.

I'm in Hampshire and can work remotely. If you're comparing a degree with a track record, the eight products are all live, so have a look and make your own mind up.